Are Indexes Unnecessary in Exadata? - Proligence

Are Indexes Unnecessary in Exadata? Arup Nanda Longtime Oracle DBA (and now DMA)

Disclaimer If you downloaded this slide deck, please note: These slides are designed merely as props to help my presentation. They are not intended to stand as independent sources of information. Therefore the contents of the slide deck are not meant to be exhaustive in any way of the content delivered at the session.

Are Indexes Unnecessary in Exadata

2

Quotes You don’t need indexes on Exadata.

Drop all the indexes and reclaim space.

Why? Because there is a storage index. Are Indexes Unnecessary in Exadata

3

3 Questions for “Best Practices”

1. Why it is better than the rest? 2. What happens if it is not followed? 3. When are they not applicable?


4

Storage Index


5

Instances and Databases Instance Combination of •Memory Areas •Background Processes

SELECT NAME FROM CUSTOMERS WHERE STATUS ='ANGRY'

datafile1 datafile2

Storage Are Indexes Unnecessary in Exadata

6

Query Processing

JILL SELECT NAME FROM CUSTOMERS WHERE STATUS ='ANGRY'

datafile1 datafile2

Database Block

Storage


7

Index

Table Are Indexes Unnecessary in Exadata

8

Components for Performance CPU Memory

Less I/O = better performance

Network I/O Controller Disk


9

The Solution • A typical query may: – Select 10% of the entire storage – Use only 1% of the data it gets

• To gain performance, the DB needs to shed weight • It has to get less from the storage  Filtering at the storage level  The storage must be cognizant of the data

CPU Memory

SELECT NAME FROM CUSTOMERS WHERE STATUS ='ANGRY'

Filtering should be Applied Here

Network I/O Controller Disk


10

CPU

iDB

Memory Network I/O Controller Disk Are Indexes Unnecessary in Exadata

11

iDB

Disk1

Disk2

Disk3


12

Disk1

SELECT … FROM TABLE WHERE COL1 = 2

Disk2

Disk3

MIN = 3

MIN = 4

MIN = 1

MIN = 3

MAX = 5

MAX = 5

MAX = 3

MAX = 5

Disk4

Storage Index Are Indexes Unnecessary in Exadata

13

Storage Indexes • Do not point to the database blocks • Merely stores for a Storage “Unit” – Max/Min Values – Whether nulls are present – For some columns

• Is on Memory of Cells; not disk – Disappears when the cell is down


14

Checking Storage Index Use select name, value/1024/1024 as stat_value from v$mystat s, v$statname n where s.statistic# = n.statistic# and n.name in ( 'cell physical IO bytes saved by storage index', 'cell physical IO interconnect bytes returned by smart scan’) Output STAT_NAME STAT_VALUE ---------- ---------SI Savings 5120.45 Smart Scan 1034.00


15

Offloading and Smart Scan

Offloading Processing to storage cells Smart Scan Reduction in I/O


16

Offloading • Column Projection select cust_id, sale_amt from sales

• Predicate Filtering where status = ‘ANGRY’

• Function Offloading select min(sale_amt)

• Virtual Columns


17

Smart Scan Benefits • Less I/O means – Faster disk access time – Less data from storage to DB – Less buffers – Less CPU – Less data between compute nodes


18

Why Not? • Pre-requisite for Smart Scan – Direct Path – Full Table or Full Index Scan – > 0 Predicates – Simple Comparison Operators

• Other Reasons – Cell is not offload capable • The diskgroup attribute cell.smart_scan_capable set to FALSE;

– Not on clustered tables, IOTs, etc. Are Indexes Unnecessary in Exadata

19

Impact of Data Distribution

SELECT … FROM TABLE WHERE COL1 = 2

MIN = 1

MIN = 1

MIN = 1

MIN = 1

MAX = 5

MAX = 5

MAX = 5

MAX = 5

MIN = 3

MIN = 4

MIN = 1

MIN = 3

MAX = 5

MAX = 5

MAX = 2

MAX = 5

Disk4


20

8 Columns C1 C2 C3 C4 C5 C6 C7 C8

C2 C3 C4 C5 C6 C7 C8 C9

C3 C4 C5 C6 C7 C8 C9 C10 Table T1


21

No Predicate • Aggregations select sum(sale_amt) from sales

Index on SALE_AMT

• Sorting select ... from sales order by sale_amt;


22

Function Based Indexes • Traditional Indexes can’t work select … from sales where to_char(sale_dt,’YY’) = ‘13’

• Function Based Indexes help • SI indexes will not be useful


23

IOTs • Index Organized Tables • PK-based rows • Secondary Indexes built on the other columns

Index

PK Are Indexes Unnecessary in Exadata

Index 24

Clustered Tables


25

Exclusion for SIs • Not for non-equality select sale_amt from sales where status != ‘SHIPPED’

• No Wildcards select sale_amt from sales where city like ‘NEW YORK%’


26

Virtual Columns • Example alter table EMP add ( tot_sal number(13) generated always as sal+comm) )

• Implication – Do not actually exists in the table – Computed at runtime


27

Indexes on Small Tables • Small table – Parameter _small_table_threshold

• Indexes still help small table http://richardfoote.wordpress.com/2009/04/16/indexes-on-small-tablespart-i-one-of-the-few/

– Less latching


28

Summary of SI Limitations • • • • • • • • •

Direct Path not used No Predicate ► No SI No Inequality (!=) ≤ 8 columns No Virtual Columns No wildcard match (LIKE ‘..%’) No IOT, Clustered Table Latching on small tables First-timer Penalty – Only subsequent queries benefit Are Indexes Unnecessary in Exadata

29

Flash Cache These are flash cards presented as disks; not memory to the Storage Cells. They are similar to SAN cache; but Oracle controls what goes on there and how long it stays. alter table person storage (cell_flash_cache keep) datafile1 datafile2

Storage Are Indexes Unnecessary in Exadata

30

Flash Trick for Indexes • Pin Oft-Used Objects in Flash SQL> alter index in_t2 storage (cell_flash_cache keep);

• Check flash CellCLI> > > >

list flashcachecontent attributes – cachedKeepSize, cachedSize, hitCount, hoursToExpiration, missCount where objectnumber = 382380;

• Or, partitions


31

Drop the Index? • Make the indexes invisible SQL> alter index i1 invisible;

– Maintains the index; but optimizer ignores it

• See the performance impact. • Selectively see the impact SQL> alter session set optimizer_use_invisible_indexes = true;

• See the performance impact.


32

Disable • Two parameters – Could be session level

• To disable offloading cell_offload_processing = false;

• To disable storage indexes alone _kcfis_storageidx_disabled = true;


33

In Conclusion • Full table scans in Exadata – may be faster compared to non-Exadata – may not be faster than index scans in Exadata – may benefit from Storage Indexes

• Storage Indexes are not same as DB Indexes • No DB Indexes helps in some cases – But not all

• Test by making DB Indexes invisible • Force FTS in those cases where index hurts Are Indexes Unnecessary in Exadata

34

Thank You! Blog: arup.blogspot.com | Twitter: arupnanda


35